1. Clinical Trials and Safety Data – Different View of the same Reality Uwe Trinks, Ph.D. PRISM Forum, San Diego

2. 2 Clinical Data Management vs. Clinical Safety Goal: Profile Drug for Use Longitudinal Data Collection Patient and Visit driven Large Part of Analysis after LPLV Mostly placebo-controlled, blinded Combined with other studies and analyzed No expedited reporting No annual reporting Data reported at the end of product development in a CTD as CDISC SDTM dataset Goal: Prevent Harm Situative Data Collection Event driven Immediate Follow up and Analysis Mostly unblinded data Expedited Reporting Annual Reporting Reporting to Ethics Committees and Investigators Reports on paper or as ICH E2BM dataset

3. 3 Efficacy vs. Safety Efficacy Safety

4. 4 Looking for the Outlayer Come on! It can‘t go wrong every time...

5. 5 Technical Hurdles

6. 6 A Study Patient Recruitment Randomization Visits FPFV LPLV Data Data Cleaning, QC, Data Management Unblinding Data Cleaning Reconciliation DB Lock Data Pooling Analysis Documentation SAE Reconciliation

7. 7 The old clinical trials paradigm Safety CTS CTMS Reconciliation Study Sites CRF SAE

8. 8 The brave new world of EDC EDC Safety CTS EDC CTMS Study Sites

9. 9 Patient View vs. Event View

10. 10 SDTM Clinical Domains by Class Interventions Events ConMeds Exposure AE MedHist Disposition Findings ECGPhysExam Labs Vitals Demog Other SubjChar Subst Use InclExcl RELREC Supp Qual Trial Design QS Comments Other Domains most relevant to safety review

11. 11 ICH ICSR DTD-XML Message Header M.1. Message Identifier Type, Format, Version, Number, Sender ID, Receiver ID, Date Case Admin Data A.1. Identification of the case safety report A.2. Primary source(s) of information A.3. Information on the sender and receiver of the case safety report Case Information B.1. Patient Characteristics B.2. Reaction(s)/Event(s) B.3. Results of tests and procedures relevant to the investigation B.4. Drug(s) information B.5. Narrative Case Summary and further information

12. 12 SDTM Clinical Domains by Class Interventions Events ConMeds Exposure AE MedHist Disposition Findings ECGPhysExam Labs Vitals Demog Other SubjChar Subst Use InclExcl RELREC Supp Qual Trial Design QS Comments Other Domains most relevant to safety review B1 B2 B3 B4

13. 13 What suits one system might not suit the next...

14. 14 Argus Interchange SAE form The ideal Process Inform Trigger Argus Safety E2B Plus file Query Dictionaries Codelists Studies Diff Report

15. 15 The real Problems

16. 16 The Problem Trigger Points EDC data collection is continuous Study Sites update information almost daily Mostly without thinking about SAE Most prominent updates: Death Dates Safety Collection is a “Snapshot” What fields and when should trigger a follow up Avoid having up to 40 FU reports Which labs and vital signs to transfer? Update of non-significant information? Feedback Loop EDC System must mark transferred data as such Requires redesign of SAE form and SAE process Requires standardization of certain form design elements

17. 17 The Problem CDISC SDTM to E2BM Conversion difficult Field inconsistencies Need to create a mapping routine (ETL tool) Easier to create the E2B file directly E2B Intake Incomplete E2B form Requires separate profile and DTD Need E2B difference report Harmonize Configuration Need to be able to automatically find Study Information in Argus Population of Suspect Drug etc. from Safety System not from EDC System Requires quick configuration of Studies

18. 18 Coding Medical Semantics Vary widely Are usually not harmonized between Safety and Clinical Coding in Safety is Real-time Most current MedDRA version Important for Reporting Coding in Clinical is Harmonized at the end Used only for Analysis Two Possibilities Harmonize Coding Strategies or Only transport Verbatim

19. 19 Major Changes in Inform Standardized SAE form Zero deviation allowed Additional fields for Lab Data Vital Signs Event Assessments Dechallenge/Rechallenge Selection Tables for Data that are uploaded by relevance Concomitant medications Medical History Vital Signs Patient Demographics Treatments

20. 20 Major Changes in Inform Pop ups in other Inform Sections “Is this an SAE/Update to SAE?” Need to be “cloned” for EVERY database Death Date Cause of Death Relatedness -> Update SAE form Other Seriousness Criteria Change in Relatedness Change in Drug Exposure Dechallenge/Rechallenge Pregnancy System Down: Fax to Safety, but enter into EDC System

21. 21 Major Changes in Inform Standardized SAE Query form Assure that answers end up in SAE form Lab Data Integration difficult Standardized and harmonized Codelists Biggest source of errors Need full alignment of CRF writers Can probably build an E2B check into Inform Trigger Mechanism Difficult balance between Regulatory Requirements Operational Efficiency Basic rule: No more than one FU per day Scheduled upload of non significant information

22. 22 Major Changes in Argus Study Information Needs to be configured before FPFV Needs to be harmonized with Clinical E2B Import Module Needs to be configured for partial E2B files Added non E2B fields Sequence numbers Event Assessments Configure for Always Accept Configure to not overwrite certain fields (e.g unblinded product information) Configure not to upload certain events (e.g entry criteria) No changes to Argus SAEs Except Narrative and Unblinding All Changes/Queries through EDC System

23. 23 Communication Most difficult task Lots of Clinical and IT lingo needs to be translated into Safety Lingo Lots of Safety lingo needs to be translated into Clinical and IT Lingo Very few people are bi- or tri-lingual

24. 24 E2B Transport Non native DTD Partial E2B form Header of non-standard format No study product information Only Study # information No Narrative Information Added fields Sequence numbers Event Assessments/Causality Non overwrite information No real acknowledgement Send Argus Case number back to Inform for sequencing and FU

25. 25 The good news It works very well at some companies GSK Technical problems are mostly solvable It saves time and money once it works

26. 26 The bad news It takes an effort to harmonize things SAE forms Codelists Dictionaries and Coding Conventions It takes a while to train people Safety to use Inform for Queries and Updates Investigators to “think safety” It requires adherence to agreed standards The “extra codelist item” will break the process Safety has to conform with not updating data in Argus Savings will happen late in the process Once the majority of the studies is in Inform and using the standard SAE process Can’t change studies midway

27. 27 Don’t forget